How a bullet actually flies

Once the bullet leaves the muzzle, the physics is deceptively simple to state and endlessly subtle in the details. Two forces dominate: gravity pulls it down, and drag slows it. Integrate those over a second or so of flight and you have most of a trajectory. But the last few inches at long range — the difference between a hit and a clean miss — live in a handful of smaller effects that a serious model layers on top: how the spin stabilizes the bullet, how that spin nudges it sideways, how a crosswind kicks it vertically, and how the Earth’s own rotation bends the path. This piece builds the trajectory from the core outward, and ends with the most honest tool of all — correcting the model to your own firearm.

The core: a point mass falling and slowing

The base model is a point-mass, three-degree-of-freedom (3-DOF) trajectory. “Point mass” means we treat the bullet as a single point with a position and a velocity, tracking where its center of mass goes without modeling how it tumbles or points. “Three degrees of freedom” are the three directions it can move: downrange, vertical, and horizontal. The governing equation is the one from the drag article:

$\frac{d\vec{v}}{dt} = -\,k\,\rho\,C_d(M)\,\lvert\vec{v}\rvert\,\vec{v} \;+\; \vec{g}$

In words: the bullet’s acceleration is drag (the first term — pointing backward along its velocity $\vec v$, scaling with air density $\rho$, the drag coefficient $C_d$ at the current Mach number, and the square of speed) plus gravity $\vec g$ (pulling straight down). There is no closed-form answer — drag depends on speed, speed changes every instant — so the model integrates numerically, stepping the bullet forward in tiny time increments with a fourth-order Runge–Kutta scheme, a standard, accurate method for exactly this kind of problem. Air density and the speed of sound come from the temperature, pressure, and humidity on the day (or a single density altitude), so Mach number and drag are right for the conditions, not a generic table.

Zeroing is solved by finding the launch angle that puts the bullet on the line of sight at your chosen zero range — a root-finder on the height-over-line-of-sight. Crosswind deflection comes from comparing the real trajectory to a drag-free one: wind matters because drag couples the bullet to a moving air mass, and the deflection is the accumulated lag.

That core gets you drop and wind to a good approximation. Now the corrections — each engaged only when its inputs are present, and labeled when it’s switched off, so you always know what the model did and didn’t account for.

Gyroscopic stability — is the bullet even flying straight?

A bullet is spun by the rifling so it flies point-first instead of tumbling, exactly like a thrown football or a spinning top. Whether the spin is enough is captured by the gyroscopic stability factor, $S_g$. The intuition of the Miller stability rule (2005) is a balance:

$S_g \;\propto\; \frac{(\text{spin rate})^2}{\text{overturning aerodynamic torque}}$

The spin rate is set by the barrel’s twist and the muzzle velocity; the overturning torque is the air trying to flip a long, fast bullet end-over-end. The reading is stark:

• $S_g < 1.0$ — unstable. The bullet tumbles. It will not fly true no matter how good everything else is.
• $S_g \approx 1.0\text{–}1.5$ — marginal; the bullet flies, but its effective ballistic coefficient is degraded.
• $S_g \gtrsim 1.5$ — comfortably stable.

Two things about where stability is worst. First, the danger point is the muzzle: $S_g$ actually grows downrange (spin decays much more slowly than forward speed), so a bullet that clears the muzzle stable only gets more stable — which is why the number is evaluated at the muzzle. Second, it depends on air density: a load that is fine in warm summer air can go marginal on a cold morning, when denser air raises the overturning torque. The model accounts for both.

Spin drift — the price of the spin

The same spin that stabilizes the bullet also steers it slightly sideways. A gyroscope under a steady toppling torque doesn’t fall over — it precesses, drifting at right angles to the push. For a bullet, the result is a slow, steady sideways walk called spin drift, growing with time of flight. Its direction is set by the twist: a right-hand twist drifts the bullet right; a left-hand twist drifts it left. It is computed from the stability factor and the flight time, and at long range it can be several inches — small, but real, and on the same order as effects shooters chase obsessively.

Aerodynamic jump — the vertical kick from a horizontal wind

Here is a genuinely counterintuitive one. A crosswind — blowing horizontally — produces a small vertical shift at the muzzle. As the bullet exits, a sideways wind briefly tips the spinning bullet’s nose, and the gyroscopic response converts that into a tiny vertical impulse imparted essentially at launch. It is called aerodynamic jump. With a right-twist barrel, a wind from the left nudges the group up; a wind from the right, down (the sense flips with twist hand). It is small, but at distance it is part of why a pure horizontal-wind correction never fully explains where the group went.

Coriolis — the Earth turns underneath the shot

Over a long flight the Earth itself rotates beneath the bullet, so a path that is “straight” in space lands slightly off from where the ground moved to. This is the Coriolis effect, and it has two parts:

• Horizontal deflection, which depends on your latitude. In the Northern Hemisphere it deflects the bullet to the right of the line of fire; in the Southern Hemisphere, to the left. It does not depend on which way you’re shooting.
• The vertical Eötvös term, which depends on latitude and firing azimuth (the compass direction you shoot). Firing east makes the bullet land slightly high; firing west, slightly low — because you are adding to or subtracting from the Earth’s eastward surface speed.

Both are small and only matter at long range, which is why they engage only when you supply latitude (and azimuth for the Eötvös part) — and are labeled off when you don’t.

Where every model is least trustworthy: transonic

Notice how many of these effects fight you at the far end of the shot — and that is exactly where the transonic region sits, roughly Mach 1.2 down to 0.9, as the bullet decelerates through the speed of sound. Here the airflow around the bullet is a shifting, shock-laden mess; the drag coefficient spikes and is hard to pin down; and the neat point-mass-plus-corrections picture is at its weakest. Any point-mass model with a standard drag function is least certain here — this is not a flaw peculiar to one tool, it is physics. An honest trajectory table flags the rows in this zone rather than presenting them with false confidence. (A full 6-DOF rigid-body simulation models the bullet’s actual tumbling and pointing and does better here — but it needs data most shooters will never have, and the point-mass model with these corrections is the honest, practical choice for the field.)

Truing: because a model predicts and a firearm decides

Every correction above is still a prediction. Your firearm is the final authority, and the honest move is to correct the model to what your bullet actually does — your dope. This is truing, and the key is knowing which knob to turn, because turning the wrong one hides an error instead of fixing it:

• Inside the supersonic range, the dominant error is almost always the muzzle velocity. So you true the MV — adjust the starting velocity until the model matches your observed drop.
• In the transonic zone and beyond, the drag model is the weak link, so you true a drag scale instead — nudging the drag curve to match your far dope.

The bullet’s Mach number at the range you’re truing on decides which knob is the honest one to touch. And there is a hard ranking underneath it: a measured chronograph mean always outranks a trued muzzle velocity. Truing is for correcting the model to reality, never for inventing a velocity you never measured — the same measured beats fitted beats estimated hierarchy that governs the whole system.

The band, not the line

One last honesty note. If you give the model a measured velocity standard deviation, it reports a ±1 SD vertical band at each range — how much your groups will spread vertically from velocity variation alone. That band is a floor on the group, not a ceiling: it is the dispersion you’ll get even if everything else is perfect. Wind, your hold, and the transonic uncertainty all live on top of it. A model that drew a single crisp line at 1000 yards would be flattering you; the honest picture is a band, and it is drawn from what you measured.